title: "Mistral NeMo | Mistral AI"

Mistral AI has released Mistral NeMo, a 12B parameter model developed in collaboration with NVIDIA that features a 128k token context window. The model, released under an Apache 2.0 license, demonstrates state-of-the-art performance for its size, outperforming models like Gemma 2 9B and Llama 3 8B on various benchmarks. It is designed for global, multilingual applications and introduces a new, more efficient tokenizer called Tekken, which improves text and code compression. Both base and instruction-tuned versions are available, with the latter showing enhanced abilities in reasoning, multi-turn conversation, and code generation.